Semantic WordRank: Generating Finer Single-Document Summarizations

TL;DR

Semantic WordRank (SWR) is an unsupervised extractive summarization method that leverages semantic and co-occurrence graphs, article structure, and topic diversity to produce superior single-document summaries.

Contribution

Introduces SWR, a novel unsupervised approach combining semantic graphs, PageRank, and spectral clustering for improved extractive summarization.

Findings

SWR outperforms state-of-the-art algorithms on DUC-02 dataset.

SWR surpasses human annotators on SummBank dataset.

SWR generates higher ROUGE scores than competing methods.

Abstract

We present Semantic WordRank (SWR), an unsupervised method for generating an extractive summary of a single document. Built on a weighted word graph with semantic and co-occurrence edges, SWR scores sentences using an article-structure-biased PageRank algorithm with a Softplus function adjustment, and promotes topic diversity using spectral subtopic clustering under the Word-Movers-Distance metric. We evaluate SWR on the DUC-02 and SummBank datasets and show that SWR produces better summaries than the state-of-the-art algorithms over DUC-02 under common ROUGE measures. We then show that, under the same measures over SummBank, SWR outperforms each of the three human annotators (aka. judges) and compares favorably with the combined performance of all judges.